M.V.P.T.Lakshika, H.A.Caldera, W.V.Welgama 301
20
th
International Conference on Advances in ICT for Emerging Regions (ICTer 2020) 05
th
– 06
th
November 2020
for decision making process and discovering correlations
between multidimensional features in the text documents [14],
[15].
III. RESEARCH PURPOSE
We are proposing a novel approach to bridge the knowledge
gap between Natural Language Processing and Data Mining
fields to generate more cohesive, readable abstractive web
news summarization using the knowledge graphs. This
approach overcome the downsides in existing abstractive
summary generation and enrich the accuracy of sentence
ranking function using the knowledge derived from
association rules in data mining for generating a better
abstractive summary. According to the literature, any existing
applications do not generate abstractive summaries on
multiple dimensions (topics) in news articles and also do not
generate ‘Updated news summaries’ for an existing
abstractive summaries. The proposed approach generate both
abstractive summaries on multiple dimensions or topics and
update summaries to help readers to read and track news
updates very easily.
IV. PROPOSED METHODOLOGY
As shown in the figure 1, after the text pre-processing steps,
ARM is applied to the automated indexes generated from
phase 1, and derived all the significant association rules from
the phase 2. In the phase 3, knowledge graph is formed with
two layers. Domain specifics in the newspapers are captured
in the data layer and semantic layer adds rich and explicit
semantics on top of the data layer to infer additional
knowledge using ontologies and lexical database such as
WordNet. A novel ranking algorithm which includes
linguistic and semantic features in the news documents along
with the knowledge mined from association rules in the phase
2 will be used to select the top-k entities from the knowledge
graph. In the phase 4, abstractive summaries are generated
using the SimpleNLG language generator. Updated
abstractive summaries are generated for an existing
abstractive summary. Generated correlations between
multidimensional features in the news documents are used to
generate multidimensional abstractive summaries. The
collection of news documents released by Document
Understanding Conference (DUC) will be used as the
experimental dataset. In the phase 5, human evaluation using
the domain experts will be carried along with the intrinsic
evaluation techniques such as recall, precision, F1 measure,
Pyramid and ROUGE evaluation.
REFERENCE
[1] R. Chettri, “Automatic Text Summarization,” Int. J. Comput. Appl.,
vol. 161, no. 1, p. 3, 2017.
[2] N. Bhatia and A. Jaiswal, “Trends in Extractive and Abstractive
Techniques in Text Summarization,” 2015, doi: 10.5120/20559-2947.
[3] A. Patil, “Automatic Text Summarization,” Int. J. Comput. Appl., vol.
109, no. 17, p. 2, 2015.
[4] N. R. Kasture, N. Yargal, N. Singh, N. N. Kulkarni, V. Mathur, “A
Survey on Methods of Abstractive Text Summarization,” 2014.
[5] D. K. Gaikwad and C. N. Mahender, “A Review Paper on Text
Summarization,” vol. 5, no. 3, p. 7, 2016.
[6] C. S. Yadav, A. Sharan, and M. L. Joshi, “Semantic graph based
approach for text mining,” in 2014 International Conference on Issues
and Challenges in Intelligent Computing Techniques (ICICT),
Ghaziabad, India, 2014, pp. 596–601, doi:
10.1109/ICICICT.2014.6781348.
[7] S. Gupta and S. K. Gupta, “Abstractive summarization: An overview
of the state of the art,” Expert Syst. Appl., vol. 121, pp. 49–65, May
2019, doi: 10.1016/j.eswa.2018.12.011.
[8] P. Kouris, G. Alexandridis, and A. Stafylopatis, “Abstractive Text
Summarization Based on Deep Learning and Semantic Content
Generalization,” in Proceedings of the 57th Annual Meeting of the
Association for Computational Linguistics, Florence, Italy, 2019, pp.
5082–5092, doi: 10.18653/v1/P19-1501.
[9] Z. H. Ali, “Multi-Document Text Summarization using Fuzzy Logic
and Association Rule Mining”, University of Technology –
Department of Computer Science, no. 41, p. 19, 2018.
[10] P. Wu, Q. Zhou, Z. Lei, W. Qiu, and X. Li, “Template Oriented Text
Summarization via Knowledge Graph,” in 2018 International
Conference on Audio, Language and Image Processing (ICALIP),
Shanghai, Jul. 2018, pp. 79–83, doi: 10.1109/ICALIP.2018.8455241.
[11] K. Gunaratna, “Semantics-based Summarization of Entities in
Knowledge Graphs,” p. 146, 2017.
[12] M. Subramaniam and V. Dalal, “Test Model for Rich Semantic Graph
Representation for Hindi Text using Abstractive Method.,” vol. 02, no.
02, p. 4, 2015.
[13] I. F. Moawad and M. Aref, “Semantic graph reduction approach for
abstractive Text Summarization,” in 2012 Seventh International
Conference on Computer Engineering & Systems (ICCES), Cairo,
Egypt, 2012, pp. 132–138, doi: 10.1109/ICCES.2012.6408498.
[14] C. E. Crangle, “Text Summarization in Data Mining,” in Soft-Ware
2002: Computing in an Imperfect World, vol. 2311, D. Bustard, W.
Liu, and R. Sterritt, Eds. Berlin, Heidelberg: Springer Berlin
Heidelberg, 2002, pp. 332–347.
[15] H. Mahgoub, D. Rösner, N. Ismail, and F. Torkey, “A Text Mining
Technique Using Association Rules Extraction,” vol. 2, no. 6, p. 8,
2008.
Authorized licensed use limited to: Hong Kong University of Science and Technology. Downloaded on July 19,2023 at 08:30:24 UTC from IEEE Xplore. Restrictions apply.